Spatial Samplings Research Articles

Real-time dual-parameter full-waveform inversion of GPR data based on robust deep learning

SUMMARY Ground penetrating radar (GPR) is becoming an increasingly important tool for understanding the shallow electrical structures of the Earth and planets due to its adaptability to harsh detection environments, efficient data acquisition and accurate detection results. GPR full-waveform can simultaneously constrain the permittivity and resistivity of the medium, providing more comprehensive geophysical information and reducing the non-uniqueness of inversion. However, given the highly non-linear inverse problem and the massive data resulted from high temporal and spatial samplings, traditional full-waveform inversion algorithms are prohibitively costly. Inspired by Google's vision semantic segmentation system, we develop a robust deep learning-guided network that integrates geology and geophysics knowledge to support the real-time translation of zero-offset GPR data into dual-parameter electrical structures. We test our proposed network using synthetic data, which demonstrates that the algorithm can provide an accurate dual-parameter electrical model from a GPR sounding in milliseconds on a common laptop PC, exhibiting high robustness and adaptability to noise interference and extreme values of model parameters. We also apply our network to field data gathered for pollutant investigation in the United States. The resulting dual-parameter structure provides a more comprehensive and realistic depiction of subsurface electrical properties and reveals the migration and ageing of pollutants. Our algorithm's real-time and accurate advantages are expected to further unleash the potential of GPR technology and enable it to play a more significant role in earth and planetary exploration.

Hydraulic fracturing distributed acoustic sensing monitoring data source mechanism inversion: A Hessian-based method

Distributed acoustic sensing (DAS) microseismic monitoring during hydraulic fracturing provides microseismic data with high spatial samplings for fracturing zones. However, sources located perpendicular to the single horizontal well of the DAS acquisition system have limited source-receiver geometries with extremely poor azimuthal coverage, resulting in high uncertainty in source mechanism inversion. To address this problem, we introduce the Hessian matrix, which governs the blurring effect caused by the source-receiver geometry, into the DAS microseismic source mechanism inversion. Our Hessian-based source mechanism inversion method consists of three main steps: (1) construct the Hessian matrix of the source mechanism based on the source-receiver geometry, (2) obtain an initial source mechanism using a conventional source mechanism inversion method, and (3) update the initial source mechanism using a Hessian-based [Formula: see text]-regularized least-squares algorithm. We assess the robustness of our method using synthetic DAS microseismic data with the consideration of noise, source location error, and different regularization parameters, and we compare the results with those of the conventional method. The results demonstrate that the Hessian-based method has a remarkable ability to mitigate the blurring effect of the Hessian matrix caused by limited DAS source-receiver geometry with poor azimuthal coverage, thereby reducing the uncertainty of the inverted source mechanism even in the presence of real noise and/or source location error. Finally, we use our method to invert the source mechanism of a real DAS microseismic event acquired during hydraulic fracturing. Our Hessian-based method provides low uncertainty in the source mechanism inversion of the real DAS microseismic data.

Extreme Events Prediction from Nonlocal Partial Information in a Spatiotemporally Chaotic Microcavity Laser.

The forecasting of high-dimensional, spatiotemporal nonlinear systems has made tremendous progress with the advent of model-free machine learning techniques. However, in real systems it is not always possible to have all the information needed; only partial information is available for learning and forecasting. This can be due to insufficient temporal or spatial samplings, to inaccessible variables, or to noisy training data. Here, we show that it is nevertheless possible to forecast extreme event occurrences in incomplete experimental recordings from a spatiotemporally chaotic microcavity laser using reservoir computing. Selecting regions of maximum transfer entropy, we show that it is possible to get higher forecasting accuracy using nonlocal data vs local data, thus allowing greater warning times of at least twice the time horizon predicted from the nonlinear local Lyapunov exponent.

Visible and short-wave infrared fiber-based snapshot imaging spectrometer with a custom high-throughput relay system

This paper presents the design and fabrication of a fiber-based snapshot imaging spectrometer working in both visible (490 nm-732 nm) and short-wave infrared (1090 nm - 1310 nm) ranges. To maximize the light collection efficiency, a custom relay system with 0.25 NA and 20 mm field of view (FOV) was designed and integrated. The bench setup showed that the custom relay system could fully resolve 10 µm fiber cores over the entire FOV among visible and short-wave infrared ranges. The numerical aperture (NA) match provided a 2.07X fold throughout improvement in the visible range and about 10X fold in the SWIR range compared to the previous generations, enabling imaging with a fast frame rate and under low illumination conditions. The presented imaging spectrometer generated spectral datacubes with 35000 spatial samplings and 23 spectral channels. Spectral urban imaging results obtained by the spectrometer in both visible and SWIR ranges are presented. Finally, we collected spectral images of apple bruising to show potential applications in the food quality industry.

Calibration-based abundances in the interstellar gas of galaxies from slit and IFU spectra

In this work, we make use of available integral field unit (IFU) spectroscopy and slit spectra of several nearby galaxies. The pre-existing empirical R and S calibrations for abundance determinations are constructed using a sample of H II regions with high-quality slit spectra. In this paper, we test the applicability of those calibrations to the IFU spectra. We estimate the calibration-based abundances obtained using both the IFU and the slit spectroscopy for eight nearby galaxies. The median values of the slit and IFU spectra-based abundances in bins of 0.1 in fractional radius Rg (normalised to the optical radius R25) of a galaxy are determined and compared. We find that the IFU and the slit spectra-based abundances obtained through the R calibration are close to each other; the mean value of the differences of abundances is 0.005 dex, and the scatter in the differences is 0.037 dex for 38 datapoints. The S calibration can produce systematically underestimated values of the IFU spectra-based abundances at high metallicities (12 + log(O/H) ≳8.55); the mean value of the differences is −0.059 dex for 21 datapoints, while at lower metallicities the mean value of the differences is −0.018 dex and the scatter is 0.045 dex for 36 data points. This provides evidence that the R calibration produces more consistent abundance estimations between the slit and the IFU spectra than the S calibration. We find that the same calibration can produce close estimations of the abundances using IFU spectra obtained with different spatial resolution and different spatial samplings. This is in line with the recent finding that the contribution of the diffuse ionised gas to the large-aperture spectra of H II regions has a secondary effect.

Study of a Shack-Hartmann wavefront sensor with adjustable spatial sampling based on spherical reference wave

For the astronomical adaptive optics systems, the observation condition is constantly changing or in a regime of low photon flux. Besides, for various brightness seeing conditions, the Shack-Hartmann wavefront sensor has a matching spatial sampling that avoids extra wavefront reconstruction error. To integrate advantageous spatial samplings, we propose a novel Shack-Hartmann wavefront sensor with adjustable spatial sampling (ASS-WFS) based on spherical reference wave. In principle, only a simple movement of the microlens array (MLA) and the CCD detector in the direction of the optical axis is required to achieve variable spatial sampling. This is also obviously advantageous in the adaptive optics performance as it allows the wavefront sensor to dynamically match brightness of the observation. To demonstrate the performance of the ASS-WFS, we present the sensor in terms of the principles, the simulation, and the experiment. Results show that the ASS-WFS is able to precisely measure and reconstruct the wavefront, and also demonstrate that advantages resulting from adjusting the spatial samplings can efficiently reduce the disadvantages of low signal-to-noise ratio to boost wavefront reconstruction accuracy.

Accurate hyperspectral imaging of mineralised outcrops: An example from lithium-bearing pegmatites at Uis, Namibia

Efficient, socially acceptable and rapid methods of exploration are required to discover new deposits and enable the green energy transition. Sustainable exploration requires a combination of innovative thinking and new technologies. Hyperspectral imaging (HSI) is a rapidly developing technology and allows for fast and systematic mineral mapping, facilitating exploration of the Earth's surface at various scales on a variety of platforms. Newly available sensors allow data capture over a wide spectral range, and provide information about the abundance and spatial location of ore and pathfinder minerals in drill-core, hand samples and outcrops with mm to cm precision. Conversely, the complex geometries of the imaged surfaces affect the spectral quality and signal-to-noise ratio (SnR) of HSI data at these very narrow spatial samplings. Additionally, the complex mineral assemblages found in hydrothermally altered ore deposits can make interpretation of spectral results a challenge. In this contribution, we propose an innovative approach that integrates multiple sensors and scales of data acquisition to help disentangle complex mineralogy associated with lithium and tin mineralisation in the Uis pegmatite complex, Namibia. We train this method using hand samples and finally produce a three-dimensional (3D) point cloud for mapping lithium mineralisation in the open pit. We were able to identify and map lithium-bearing cookeite and montebrasite at outcrop scale. The accuracy of the approach was validated by drill-core data, XRD analysis and LIBS measurements. This approach facilitates efficient mapping of complex terrains, as well as important monitoring and optimisation of ore extraction. Our method can easily be adapted to other minerals relevant to the mining industry.

Combining laboratory measurements and proximal soil sensing data in digital soil mapping approaches

Digital soil mapping (DSM) products are limited in accuracy because of the lack of soil inputs. Soil sensing is a promising alternative to direct soil measurements that could provide much denser spatial samplings. Although using relevant and detailed soil sensing input in DSM is considered as vital to increase the prediction performances, there has been no studies in the literature that compare and develop the methods for integrating new sources of soil data that can be applied as inputs of DSM. This paper fills this gap on the example of mapping electrical conductivities from sites with laboratory measurements, in-field EM38MK2 measurements and spatially exhaustive covariates. Three different approaches are tested for putting in synergy real measurements and EM38MK2 measurements: (i) EM38MK2 measurement considered as measured points, (ii) EM38MK2 measurement used for building a new soil covariate and (iii) EM38MK2 measurement considered as a soft data in a regression co-kriging approach. According to soil analysis's financial expenditure, choosing an optimal sample size to merge laboratory analysis and in-field EM38MK2 measurements as surrogate data was done on the best method. The results showed (i) the utility of EM38MK2 data in DSM as a surrogate input data for mapping soil salinity (ii) Regression co-kriging was the best method for integration and (iii) The impact of EM38MK2 data on the gains of performance becomes greater and greater as the sizes of real measurements of soil salinity decrease. Hence, in other areas worldwide that soil sensing as alternative data is accessible, this research's future utilization could be possible as a promising way to tackle one of the essential constraints of DSM.

IMRT QA and gamma comparisons: The impact of detector geometry, spatial sampling, and delivery technique on gamma comparison sensitivity.

To separately quantify sensitivity differences in patient-specific quality assurance comparisons analyzed with the gamma comparison for different measurement geometries, spatial samplings, and delivery techniques [intensity modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT)]. Error-free calculations for 20 IMRT and 20 VMAT cases were compared to calculations with known induced errors of varying magnitudes, using gamma comparisons. Five error types (MU scaling, three different MLC errors, and collimator errors) were induced in plan calculations on three different detector geometries - ArcCHECK, MapCHECK, and Delta 4. To study detector geometry sensitivity effects alone, gamma comparisons were made with 1mm error-free calculations compared to 1mm error-induced calculations for each device. Effects of spatial sampling were studied by making the same gamma comparisons, but down-sampling the error-induced calculations to the real spatial sampling of each device. Additionally, 1mm vs 1mm comparisons between the IMRT and VMAT cases were compared to investigate sensitivity differences between IMRT and VMAT using IMRT and VMAT cohorts with similar ranges of plan complexity and average aperture size. For each case, induced error type, and device, five different gamma criteria were studied to ensure sensitivity differences between devices, spatial sampling scenarios, and delivery technique were not gamma criterion specific, resulting in over 36,000 gamma comparisons. For IMRT cases, Delta4 and MapCHECK devices had similar error sensitivities for lagging leaf, bank shift, and MU errors, while the ArcCHECK had considerably lower sensitivity than the planar-type devices. For collimator errors and perturbational leaf errors the ArcCHECK had higher error sensitivity than planar-type devices. This behavior was independent of gamma parameters (percent dose difference, distance-to-agreement, and low dose threshold), though use of local normalization resulted in error sensitivites that were markedly similar between all three devices. Differences between detector geometries were less pronounced for VMAT deliveries. Error sensitivity for a given gamma criterion when comparing IMRT and VMAT deliveries on the same devices showed that VMAT plans were more sensitive to some specific error types and less sensitive to others, when compared to IMRT plans. For the ArcCHECK device, the sensitivity of IMRT and VMAT cases was quite similar, whereas this was not the case for the planar-type devices. When comparing error sensitivity between 1mm vs 1mm calculations and 1mm vs the real spatial sampling for each device, results showed that increased spatial sampling did not systematically increase error sensitivity. Noticeable differences in error sensitivity were observed for different detector geometries, but differences were dependent on induced error type, and a particular device geometry did not offer universal improvements in error sensitivity across studied error types. This study demonstrates that the sensitivity of the gamma comparison does not largely hinge on detector spatial sampling. VMAT deliveries were generally less sensitive to errors when compared to IMRT plans for the planar-type devices, while similar sensitivities were observed between delivery techniques for the ArcCHECK device. Results of this work suggest that a universal gamma criterion is inappropriate for IMRT QA and that the percent pixels passing is an insufficient metric for evaluating quality assurance checks in the clinic.

An EnKF-based method to produce rainfall maps from simulated satellite-to-ground MW-link signal attenuation.

Measuring rainfall is complex, due to the high temporal and spatial variability of precipitation, especially in a changing climate, but it is of great importance for all the scientific and operational disciplines dealing with rainfall effects on the environment, human activities, and economy. Microwave (MW) telecommunication links carry information on rainfall rates along their path, through signal attenuation caused by raindrops, and can become measurements of opportunity, offering inexpensive chances to augment information without deploying additional infrastructures, at the cost of some smart processing. Processing satellite telecom signals bring some specific complexities related to the effects of rainfall boundaries, melting layer, and non-weather attenuations, but with the potential to provide worldwide precipitation data with high temporal and spatial samplings. These measurements have to be processed according to the probabilistic nature of the information they carry. An EnKF-based (Ensemble Kalman Filter) method has been developed to dynamically retrieve rainfall fields in gridded domains, which manages such probabilistic information and exploits the high sampling rate of measurements. The paper presents the EnKF method with some representative tests from synthetic 3D experiments. Ancillary data are assumed as from worldwide-available operational meteorological satellites and models, for advection, initial and boundary conditions, rain height. The method reproduces rainfall structures and quantities in a correct way, and also manages possible link outages. It results computationally viable also for operational implementation and applicable to different link observation geometries and characteristics.

Pragmatic spatial sampling for wearable MEG arrays

Several new technologies have emerged promising new Magnetoencephalography (MEG) systems in which the sensors can be placed close to the scalp. One such technology, Optically Pumped MEG (OP-MEG) allows for a scalp mounted system that provides measurements within millimetres of the scalp surface. A question that arises in developing on-scalp systems is: how many sensors are necessary to achieve adequate performance/spatial discrimination? There are many factors to consider in answering this question such as the signal to noise ratio (SNR), the locations and depths of the sources, density of spatial sampling, sensor gain errors (due to interference, subject movement, cross-talk, etc.) and, of course, the desired spatial discrimination. In this paper, we provide simulations which show the impact these factors have on designing sensor arrays for wearable MEG. While OP-MEG has the potential to provide high information content at dense spatial samplings, we find that adequate spatial discrimination of sources (< 1 cm) can be achieved with relatively few sensors (< 100) at coarse spatial samplings (~ 30 mm) at high SNR. After this point approximately 50 more sensors are required for every 1 mm improvement in spatial discrimination. Comparable discrimination for traditional cryogenic systems require more channels by these same metrics. We also show that sensor gain errors have the greatest impact on discrimination between deep sources at high SNR. Finally, we also examine the limitation that aliasing due to undersampling has on the effective SNR of on-scalp sensors.

Modelling Reservoir Chlorophyll-a, TSS, and Turbidity Using Sentinel-2A MSI and Landsat-8 OLI Satellite Sensors with Empirical Multivariate Regression

Sentinel-2A/MSI (S2A) and Landsat-8/OLI (L8) data products present a new frontier for the assessment and retrieval of optically active water quality parameters including chlorophyll-a (Chl-a), suspended particulate matter (TSS), and turbidity in reservoirs. However, because of their differences in spatial and spectral samplings, it is critical to evaluate how well the sensors are suited for the seamless generation of the water quality parameters (WQPs). This study presents results from the retrieval of the WQP in a reservoir from L8 and S2A optical sensors, after atmospheric correction and standardization through band adjustment. An empirical multivariate regression model (EMRM) algorithmic approach is proposed for the estimation of the water quality parameters in correlation with in situ laboratory measurements. From the results, both sensors estimated Chl-a concentrations with R2 of greater than 70% from the visible green band for L8 and a combination of green and SWIR-1 bands for S2A. While the NMSE% was nearly the same for both sensors in Chl-a estimation, the RMSE was &lt;10 μg/L and &gt;10 μg/L for L8 and S2A estimations of Chl-a, respectively. For TSS retrieval, L8 outperformed S2A by 31% in accuracy with R2&gt;0.9 from L8’s red, blue, and green bands, as compared to 0.47≤R2≥0.61 from S2A’s red and NIR bands. The RMSE were the same as for Chl-a, and the NMSE% were both in the same range. Both sensors retrieved turbidity with high and nearly equal accuracy of R2&gt;70% from the visible and NIR bands, with equal RMSE at &lt;10% NTU and NMAE% from S2A being higher by more than 30% as compared to L8’s NMAE% at 15%. The study concluded that the higher performance accuracy of L8 is attributed to its higher SNR and spectral bandwidth placement as compared to S2A bands. Comparatively, S2A overestimated Chl-a and turbidity but performed equally well compared to OLI in the estimation of TSS. The results show that while absolute accuracy of retrieval of the WQPs still requires improvements, the developed algorithms are broadly able to discern the biooptical water quality in reservoirs.

Analysing the impact of soil spatial sampling on the performances of Digital Soil Mapping models and their evaluation: A numerical experiment on Quantile Random Forest using clay contents obtained from Vis-NIR-SWIR hyperspectral imagery

It has long been acknowledged that the soil spatial samplings used as inputs to DSM models are strong drivers – and often limiting factors – of the performances of such models. However, few studies have focused on evaluating this impact and identifying the related spatial sampling characteristics. In this study, a numerical experiment was conducted on this topic using the pseudo values of topsoil clay content obtained from an airborne Visible Near InfraRed-Short Wave InfraRed (Vis-NIR-SWIR) hyperspectral image in the Cap Bon region (Tunisia) as the source of the spatial sampling.Twelve thousand DSM models were built by running a Random Forest algorithm from soil spatial sampling of different sizes and average spacings (from 200 m to 2000 m) and different spatial distributions (from clustered to regularly distributed), aiming to mimic the various situations encountered when handling legacy data. These DSM models were evaluated with regard to both their prediction performances and their ability to estimate their overall and local uncertainties. Three evaluation methods were applied: a model-based one, a classical model-free one using 25% of the sites removed from the initial soil data, and a reference one using a set of 100,000 independent sites selected by stratified random sampling over the entire region.The results showed that: 1) While, as expected, the performances of the DSM models increased when the spacing of the sample increased, this increase was diminished for the smallest spacing as soon as 50% of the spatially structured variance was captured by the sampling, 2) Sampling that provided complete and even distributions in the geographical space and had as great spread of the target soil property as possible increased the DSM performances, while complete and even sampling distributions in the covariate space had less impacts, 3) Systematic underestimations of the overall uncertainty of DSM models were observed, that were all the more important that the sparse samplings poorly covered the real distribution of the target soil property and that the dense sampling were unevenly distributed in the geographical space, 4) The local uncertainties were underestimated for sparse sampling and over-estimated for dense sampling while being sensitive to the same sampling characteristics as overall uncertainty.Such finding have practical outcomes on sampling strategies and DSM model evaluation that are discussed.

Is the recovery of stratospheric O&amp;lt;sub&amp;gt;3&amp;lt;/sub&amp;gt; speeding up in the Southern Hemisphere? An evaluation from the first IASI decadal record (2008–2017)

Abstract. In this paper, we present the global fingerprint of recent changes in middle–upper stratosphere (MUSt; &lt;25 hPa) ozone (O3) in comparison with lower stratosphere (LSt; 150–25 hPa) O3 derived from the first 10 years of the IASI/Metop-A satellite measurements (January 2008–December 2017). The IASI instrument provides vertically resolved O3 profiles with very high spatial and temporal (twice daily) samplings, allowing O3 changes to be monitored in these two regions of the stratosphere. By applying multivariate regression models with adapted geophysical proxies on daily mean O3 time series, we discriminate anthropogenic trends from various modes of natural variability, such as the El Niño–Southern Oscillation (ENSO). The representativeness of the O3 response to its natural drivers is first examined. One important finding relies on a pronounced contrast between a positive LSt O3 response to ENSO in the extratropics and a negative one in the tropics, with a delay of 3 months, which supports a stratospheric pathway for the ENSO influence on lower stratospheric and tropospheric O3. In terms of trends, we find an unequivocal O3 recovery from the available period of measurements in winter–spring at middle to high latitudes for the two stratospheric layers sounded by IASI (&gt;∼35∘ N–S in the MUSt and &gt;∼45∘ S in the LSt) as well as in the total columns at southern latitudes (&gt;∼45∘ S) where the increase reaches its maximum. These results confirm the effectiveness of the Montreal Protocol and its amendments and represent the first detection of a significant recovery of O3 concurrently in the lower, in the middle–upper stratosphere and in the total column from one single satellite dataset. A significant decline in O3 at northern mid-latitudes in the LSt is also detected, especially in winter–spring of the Northern Hemisphere. Given counteracting trends in the LSt and MUSt at these latitudes, the decline is not categorical in total O3. When freezing the regression coefficients determined for each natural driver over the whole IASI period but adjusting a trend, we calculate a significant speeding up in the O3 response to the decline of O3-depleting substances (ODSs) in the total column, in the LSt and, to a lesser extent, in the MUSt, at high southern latitudes over the year. Results also show a small significant acceleration of the O3 decline at northern mid-latitudes in the LSt and in the total column over the last few years. That, specifically, needs urgent investigation to identify its exact origin and apprehend its impact on climate change. Additional years of IASI measurements would, however, be required to confirm the O3 change rates observed in the stratospheric layers over the last few years.

Forest Canopy Height and Gaps from Multiangular BRDF, Assessed with Airborne LiDAR Data

Both vegetation multi-angular and LiDAR (light detection and ranging) remote sensing data are indirectly and directly linked with 3D vegetation structure parameters, such as the tree height and vegetation gap fraction, which are critical elements in above-ground biomass and light profiles for photosynthesis estimation. LiDAR, particularly LiDAR using waveform data, provides accurate estimates of these structural parameters but suffers from not enough spatial samplings. Structural parameters retrieved from multiangular imaging data through the inversion of physical models have larger uncertainties. This study searches for an analytical approach to fuse LiDAR and multiangular data. We explore the relationships between vegetation structure parameters derived from airborne vegetation LiDAR data and multiangular data and present a new potential angle vegetation index to retrieve the tree height and gap fraction using multi-angular data in Howland Forest, Maine. The BRDF (bidirectional reflectance distribution factor) index named NDMM (normalized difference between the maximum and minimum reflectance) linearly increases with the tree height and decreases with the gap fraction. In addition, these relationships are also dependent on the wavelength, tree species, and stand density. The NDMM index performs better in conifer (R = 0.451 for tree height and R = 0.472 for the gap fraction using the near infrared band) than in deciduous and mixed forests. It is superior in sparse (R = 0.569 for tree height and R = 0.604 for the gap fraction using the near infrared band) compared to dense forest. Moreover, the NDMM index is more strongly related to tree height and the gap fraction at the near infrared band than at the three visible bands. This study sheds light on the possibility of using multiangular data to map vegetation’s structural parameters in larger regions for carbon cycle studies through the fusion of LiDAR and multiangular remote sensing data.

GPU-based Green’s function simulations of shear waves generated by an applied acoustic radiation force in elastic and viscoelastic models

Shear wave calculations induced by an acoustic radiation force are very time-consuming on desktop computers, and high-performance graphics processing units (GPUs) achieve dramatic reductions in the computation time for these simulations. The acoustic radiation force is calculated using the fast near field method and the angular spectrum approach, and then the shear waves are calculated in parallel with Green’s functions on a GPU. This combination enables rapid evaluation of shear waves for push beams with different spatial samplings and for apertures with different f/#. Relative to shear wave simulations that evaluate the same algorithm on an Intel i7 desktop computer, a high performance nVidia GPU reduces the time required for these calculations by a factor of 45 and 700 when applied to elastic and viscoelastic shear wave simulation models, respectively. These GPU-accelerated simulations also compared to measurements in different viscoelastic phantoms, and the results are similar. For parametric evaluations and for comparisons with measured shear wave data, shear wave simulations with the Green’s function approach are ideally suited for high-performance GPUs.

The MSG-SEVIRI-based cloud property data record CLAAS-2

Abstract. Clouds play a central role in the Earth's atmosphere, and satellite observations are crucial for monitoring clouds and understanding their impact on the energy budget and water cycle. Within the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) Satellite Application Facility on Climate Monitoring (CM SAF), a new cloud property data record was derived from geostationary Meteosat Spinning Enhanced Visible and Infrared Imager (SEVIRI) measurements for the time frame 2004–2015. The resulting CLAAS-2 (CLoud property dAtAset using SEVIRI, Edition 2) data record is publicly available via the CM SAF website (https://doi.org/10.5676/EUM_SAF_CM/CLAAS/V002). In this paper we present an extensive evaluation of the CLAAS-2 cloud products, which include cloud fractional coverage, thermodynamic phase, cloud top properties, liquid/ice cloud water path and corresponding optical thickness and particle effective radius. Data validation and comparisons were performed on both level 2 (native SEVIRI grid and repeat cycle) and level 3 (daily and monthly averages and histograms) with reference datasets derived from lidar, microwave and passive imager measurements. The evaluation results show very good overall agreement with matching spatial distributions and temporal variability and small biases attributed mainly to differences in sensor characteristics, retrieval approaches, spatial and temporal samplings and viewing geometries. No major discrepancies were found. Underpinned by the good evaluation results, CLAAS-2 demonstrates that it is fit for the envisaged applications, such as process studies of the diurnal cycle of clouds and the evaluation of regional climate models. The data record is planned to be extended and updated in the future.

Minimum variance projection for direct measurements of power-law spectra in the wavenumber domain

Abstract. Minimum variance projection is widely used in geophysical and space plasma measurements to identify the wave propagation direction and the wavenumber of the wave fields. The advantage of the minimum variance projection is its ability to estimate the energy spectra directly in the wavenumber domain using only a limited number of spatial samplings. While the minimum variance projection is constructed for discrete signals in the data, we find that the minimum variance projection can reasonably reproduce the spectral slope of the power-law spectrum if the data represent continuous power-law signals. The spectral slope study using the minimum variance projection is tested against synthetic random data with a power-law spectrum. The method is applicable even for a small number of spatial samplings. Conversely, the spatial aliasing causes a flattening of the spectrum.

Applying multiple land surface temperature products to derive heat fluxes over a grassland site

Land heat fluxes are essential components of the water and energy cycle and important variables in the management of agronomy and forestry resources. The estimation of the heat fluxes can be done with a number of methodologies, with some of them having the land surface temperature (Ts) as one of their key inputs to derive the fluxes. Here the production of Ts-driven surface heat fluxes over a grassland site in Switzerland is demonstrated by running a specific heat flux methodology (SEBS) fed by a number of satellite Ts estimates (from the instruments AATSR, MODIS, SEVIRI, AMSR-E, and SSMIS). The Ts estimates are compared with an in situ estimate derived from radiometric observations at the station, and the satellite latent heat flux (LE) estimates with the station Eddy Covariance (EC) measurements. The satellite Ts products include estimates at different spatial resolutions (from ∼1 to ∼25km) and time samplings (from 2 overpasses per day to 1/2 hourly observations). Root Mean Square Differences (RMSDs) between the daytime satellite and station Ts are 2.72 (AATSR), 4.41 (MODIS), 3.59 (SEVIRI), 3.81 (AMSR-E), and 2.79 (SSMIS), but given the different time samplings and spatial resolutions its is difficult to be conclusive about the accuracy of the Ts estimates. Concerning the flux estimates, for those sensors with midday overpasses, a RMSD of ∼25% are found when comparing the instantaneous latent flux (LE, or evaporation expressed as an amount of water) at satellite overpass with the EC observations, which compares well with the accuracy reported elsewhere for similar landscapes. Given that both Ts and LE are evaluated at the station, a link between Ts and LE accuracy is investigated, but its is not apparent for this specific comparison. This could be related to SEBS accuracy also depending on other variables, apart from Ts, but also to the representativeness of the metrics used for the evaluation given the spatial miss-matches existing between the satellite estimates and station observations. Discrepancies were observed between the EC fluxes, the measured surface available energy, and the evaporation estimates from a lysimeter also present at the station, illustrating also the difficulties of the ground observations to provide accurate heat fluxes for satellite evaluation.

Mixture model-based atmospheric air mass classification: a probabilistic view of thermodynamic profiles

Abstract. Air mass classification has become an important area in synoptic climatology, simplifying the complexity of the atmosphere by dividing the atmosphere into discrete similar thermodynamic patterns. However, the constant growth of atmospheric databases in both size and complexity implies the need to develop new adaptive classifications. Here, we propose a robust unsupervised and supervised classification methodology of a large thermodynamic dataset, on a global scale and over several years, into discrete air mass groups homogeneous in both temperature and humidity that also provides underlying probability laws. Temperature and humidity at different pressure levels are aggregated into a set of cumulative distribution function (CDF) values instead of classical ones. The method is based on a Gaussian mixture model and uses the expectation–maximization (EM) algorithm to estimate the parameters of the mixture. Spatially gridded thermodynamic profiles come from ECMWF reanalyses spanning the period 2000–2009. Different aspects are investigated, such as the sensitivity of the classification process to both temporal and spatial samplings of the training dataset. Comparisons of the classifications made either by the EM algorithm or by the widely used k-means algorithm show that the former can be viewed as a generalization of the latter. Moreover, the EM algorithm delivers, for each observation, the probabilities of belonging to each class, as well as the associated uncertainty. Finally, a decision tree is proposed as a tool for interpreting the different classes, highlighting the relative importance of temperature and humidity in the classification process.